Strategy Research And Implementation Of Advertising Mapping Based On Paddlepaddle

The purpose of Commercials is to display advertising to target users,and to attract users to complete the transaction under the requirements of Advertising Law of People's Republic of China,throughout the development of Internet advertising in the past two decades,Commercials has experienced the PC era,mobile Internet era,artificial intelligence era.Today's Internet advertising is no longer a matter of simple content aggregation and distribution,but is gradually transformed into a form of precision marketing and each person sees something unique.Standing in advertiser position,it is necessary to put the right advertisements at the right time,with the correct method to users who have needs.At the same time,each user sees advertisements that are in line with the user's specific attributes such as age,income,and personality.A proper amount of commercial advertisements can not only increase the consumer's horizon,but also the user's consumer behavior.It also plays a role in aiding decision-making.Many domestic and foreign Internet companies have deep product experience and technology accumulation in the commercial advertising industry.For example,Alphat,Baidu,Discovering a large number of commercial advertising styles can be divided into four types: text ads,image ads,text&image ads,and video advertising,this article researches the PaddlePaddle,an open source platform for deep learning.It also describes in detail the process of source code compilation,installation,environment dependence,open source code and documentation,and the platform's contribution.The performance optimization used by the Python version,Using PaddlePaddle to achieve the advertising map strategy that is mentioned in this article.Through a large number of investigations and findings,the experience of the current tourism industry is more and more sought for quality and individuality.This paper proposes an innovative advertising style and introduces in detail the functions and mapping strategies of each part of the new style,combined with the PaddlePaddle.The key technology of the image classification section trains a graphic-text relevance model tailored to the advertising of the tourism industry;it replaces datasets in the text similarity calculation section of PaddlePaddle,and trains words that perform better on short texts.Useing the word vector model to classify short texts with a very high degree of similarity.Short texts of the same category have the feature of mutual replacement.At the same time,short texts of the same category can be bound to the same picture.This describes the text around the picture.Enriched,with graphic ads,through the above-mentioned advertising graphic strategy,in the process of matching pictures,it not only captures the user's pursuit of relaxation and leisure travel,but also guarantees the advertisement layout and advertisement title,advertisement layout and description text.The high relevance also guarantees the effectiveness and conversion rate of commercial advertising and promotion,The map material was evaluated and verified by experiments.Based on the experimental data analysis,it was concluded that the innovative advertising style and the proposed mapping strategy designed in this paper were significantly better than the plain text style in both click rate and conversion rate.